Forbidden Cheese Development crest Forbidden Cheese
ForbiddenReader app iconDesktop app

Forbidden Cheese Development

ForbiddenReader

Your documents, read aloud — by your own machine, in a real voice, with nothing wandering off to the cloud.

We could not train an actual bear to read your PDFs to you. This is the next best thing: neural text-to-speech that runs entirely on your computer. No account, no subscription, no data leaving your rig. Load a file, hit Speak, done.

A bear in a work shirt and boots reads a leather-bound book titled 'The Wisdom of the Old Ways' beside a blacksmith's forge, a mug marked BEAR FUEL steaming nearby.
Fig. 1 — The reader at rest. Fuel: as marked.

Chapter the First — What it does

It turns quiet pages into a voice in the room.

Feed ForbiddenReader a plain-text, Markdown, Word, or PDF file, and it converts the words to speech using Piper’s neural voices — then reads them aloud through your speakers, or saves the whole reading to a WAV you can keep. Every bit of that happens on your own computer. Nothing is uploaded. Nothing phones home. The document goes in; a voice comes out; the internet is never invited.

Chapter the Second — The clever bit

It starts reading before it’s done thinking.

Most readers make you wait while the whole document renders. This one doesn’t. Playback begins on paragraph one while the rest of the file generates quietly in the background — so a forty-page PDF starts talking almost immediately.

Under the hood, a producer keeps a few paragraphs generated and buffered ahead of a consumer that plays them, so there’s no gap between one paragraph and the next. Seamless out loud; patient underneath. Rather like reading by a fire while the next log is already catching.

Chapter the Third — The full inventory

Everything it knows how to do.

Reads four formats

txt · md · docx · pdf — loaded, cleaned, and spoken.

Gen-while-speak

Playback starts on paragraph one; the rest catches up in the background.

Save to WAV

Export a full reading as one audio file — roughly a minute per 2,600 words.

Save clean TXT

Pull the stripped, readable plain text out of any document you load.

Speed slider

Nudge the pace from 0.75× to 1.50× before you hit Speak.

Stop on a dime

Halts playback and generation cleanly, mid-sentence if you like.

Pick your voice

Any Piper voice you fancy. A Northern English gent is the recommended start.

Fully offline

Once it’s set up, it never needs the internet again. Ever.

Chapter the Fourth — The oath

No cloud. No account. No nonsense.

Built with Python and open-source tools, ForbiddenReader runs on your machine and answers to you alone. There is no telemetry to switch off, no login to forget, no subscription to cancel.

“The only thing it sends anywhere is your document — out loud, into the room.”

100% offline No subscription No account No telemetry Open source Powered by Piper TTS

Chapter the Last — Take it home

Free, on your terms.

Grab the source, follow the README, point it at Piper and a voice, and you’re reading. It wants Python 3, Piper TTS, and VLC — the setup is walked through step by step.

pip install python-docx pymupdf Pillow opencv-python python-vlc

Requires: Python 3 · Piper TTS · VLC (full install). Windows-first; the guts are plain Python.